Training pipeline failed with error message: Train set should contain all labels.

[VenkataKomaragi]

Hi Team,

   I just started evaluating Google Vertex using my sandbox . I m building a basic classification pipeline with Tabular data. I have created Data Set under us-west-1 and started creating pipeline.When submitted the model for training, it failed with "Training pipeline failed with error message: Train set should contain all labels." When I try to create labelling tasks, it is only showing us-central1 region.

Any pointers on this?

Thanks

Venkata

[lsolatorio]

Hi @VenkataKomaragi, 

Welcome and thank you for reaching out to our community for help.

This error message is usually triggered if the train split was not able to get the necessary labels due to a small dataset or multiple data duplicates. One way to resolve this is through manual data splits to ensure that all labels appear in the training split as described in this document. 

Refer to this "Data splits for tabular data" article for more information.

Hope this helps.

[felicitas_b]

I am having a similar issue - working on a small data set for an online course. So this means that the error might be due to the small set? How can I recover? I have just a simple binary task and I tried the default split with randomness but ... internal error etc.